Open Source
llama.cpp b9533 release expands platform support
llama.cpp Releaseshigh confidence
Why it matters
- →Expands platform support, enhancing compatibility for diverse systems.
- →ROCm 7.2 inclusion is significant for AMD GPU users.
- →Maintains focus on refining existing capabilities rather than new features.
The latest b9533 release of llama.cpp focuses on expanding and refining platform support. Notably, it includes Vulkan support for both Ubuntu and Windows, while maintaining CUDA support with updated DLLs for Windows. However, KleidiAI support for macOS Apple Silicon is disabled in this release. The addition of ROCm 7.2 for Ubuntu x64 is a key update for AMD GPU users, enhancing compatibility. This release underscores llama.cpp's commitment to broad platform compatibility, though it doesn't introduce new model architectures.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1llama.cpp b9494 release expands platform support
llama.cpp Releases · June 4, 2026 · Same story
- 2llama.cpp b9535 release expands platform support
llama.cpp Releases · June 7, 2026 · Same story
More from llama.cpp Releases
Coding Toolscoding
llama.cpp b12022 adds ROCm 10 and Snapdragon support
This release quietly cements llama.cpp as the universal inference runtime by finally bringing ROCm 10.0 to Linux and Windows alongside CUDA 13.4, effectively closing the hardware gap for AMD users who were previously left behind. The inclusion of native Snapdragon support on Linux marks a significant step toward running local LLMs on ARM-based mobile chips without emulation overhead. While KleidiAI on Apple Silicon is disabled in this build, the broader expansion to diverse accelerators ensures that developers aren't locked into NVIDIA ecosystems anymore.
llama.cpp Releases
Coding Toolscoding
llama.cpp b12023 fixes shared prefix bug
This release addresses a specific inference artifact where decision prefixes were incorrectly bleeding into output embeddings. By capping shared prefixes at the first output position, llama.cpp ensures that option tokens remain distinct for parent and child tasks. This is a targeted fix for multi-step reasoning or voting workflows rather than a performance boost. It stabilizes complex generation patterns without changing the underlying model architecture.
llama.cpp Releases
Coding Toolscoding
llama.cpp b12024 adds Snapdragon and ROCm 10
This release quietly expands llama.cpp's hardware support to include Qualcomm Snapdragon devices with CPU, Adreno GPU, and Hexagon NPU acceleration. It also updates ROCm to version 10.0 for both Linux and Windows, keeping AMD users in step with the latest driver stack. CUDA builds now default to 13.4 libraries alongside 12.x options, reflecting the industry's gradual shift toward newer NVIDIA toolkits. While no new model architectures are introduced, the binary matrix grows significantly, ensuring local inference remains viable on a wider array of edge and server hardware.
llama.cpp Releases
More in Open Source
© Lev SelectorOpen Sourcemodels
Mistral Large 4 Released as Open Weights Model
French AI startup Mistral has released its latest large language model, Large 4, with open weights available for download.
Lev Selector
© The Verge AIOpen Sourceother
Akai opens MPC Sample Key firmware to modders
Akai is betting that the community will build better instruments than its engineers can. By open-sourcing the firmware for the budget MPC Sample Key and providing an SDK, they are handing the keys to hackers who have already turned the device into a synth powerhouse using AI-assisted coding. This isn't just about patches; it's a strategic pivot where Akai provides the hardware canvas while the community paints the OS. It signals a shift from closed ecosystems to developer-led innovation in music tech.
The Verge AI
© Sam WitteveenOpen Sourceimage
MOSS Image: Open Diffusion Model with JSON Layout Control
OpenMOSS-Team released MOSS Image, a 5B parameter diffusion transformer that finally brings structural control to open-source image generation. Unlike standard models where text placement is a roll of the dice, this model accepts JSON layouts to dictate exactly where elements and text appear in the final render. The Apache 2 license removes legal friction for commercial use, while support for ComfyUI and SGLang makes it immediately accessible to builders. This shifts image generation from pure prompt-based chaos to deterministic composition, a critical step for practical design workflows.
Sam Witteveen